Newsgroups: comp.robotics
Path: brunix!uunet!cis.ohio-state.edu!news.sei.cmu.edu!fs7.ece.cmu.edu!crabapple.srv.cs.cmu.edu!news
From: nivek@cs.cmu.edu (Kevin Dowling)
Subject: Positioning system
Message-ID: <1992Jul09.180237.203281@cs.cmu.edu>
Date: Thu, 09 Jul 92 18:02:37 GMT
Organization: Robotics Institute, Carnegie Mellon
Nntp-Posting-Host: scythe.frc.ri.cmu.edu
Reply-To: nivek@cs.cmu.edu
Lines: 82


I have often desired a simple wireless (RF-based), high data rate, kilometer
range, centimeter accuracy positioning system and have found one that
comes close. A recent issue of Sensors had a brief article on it.

Precision Technology.  Utilizes stationary receiving antennas and
vehicle transmitter to determine transmitter (vehicle) position.
Essentially uses phase difference between received signals to
determine position within 1-10cm over several kilometer range.  Uses
40MHz VHF signal. Sample rate is 100Hz, with the software overhead
they end up with about 20Hz rates. System is not necessarily line 'o sight
either. Pretty damn good.

Currently being used for race car applications for using this position
information to derive accelerations (cetripetal, angular, bearing,
etc), velocities, inferred information includes braking points,
turn-in etc. Setup at three provings grounds, Dearborn, Naples (FL) a
race track in England. Jackie Stewart is part owner of the company as
well.

The system was actually developed in New Zealand. Systems have been
installed at Nissan, Ford and a race track in England. The company's
focus has been motorsports where holding the line through a corner
consistently can mean winning or losing a race. I think with our
(CMU's) interest they will be examining other areas as well.
(i.e. vehicle tracking for autonavigation, construction automation,
exploring etc.)

A couple of people from our lab have gone to see the system in action
at the Dearborn proving grounds. However, in a simple test they moved
an equipped T-bird a set distance and measured an error of +/- 18cm.
Pretty big. They aren`t sure why it was off but believe that Ford
had been altering the Kalman parameters for the tracking. ARC will be
getting back to us on the system.

Questions we had:

Q. Baseline: Does the system require large baselines for the touted
1-10cm accuracy over kilometers? I gave a hypothetical situation of a
10 acre parking lot for a tractor-trailer application.
A. As long as the antennas are distanced from the periphery of the
operating area by 9 meters or so no large baseline is needed. I asked
specifically about a site that is 100's of meters on a side and the
answer was still about 9 meters from the edge of the operating area.

Q. Ambiguity interval. Does it affect postion tracking as it does rangefinders?
A. Apparently not. [Not sure why yet]

Q. Effects of large steel buildings, bridge structures?
A. Can affect it. Usually answer is to expand 4 antenna system to 5 or
6 antennas to 'fix' those areas.

Q. Price?
A. Antenna receiver is $6K, Transmitter $7.5K, Cabling runs est
$350/1000' So a four antenna system and a mile of wire the cost is
around $35K. However, what ARC really wants to sell is their tracking and
display software. This can really drive the cost up into the $100K
ranges. I think for most robotics applications people will initially roll
their own software so the cost is not prohibitive.

Q. Does it return orientation?
A. No.  A separate heading sensor will
have to be used. Although it you have steering angle, direction of
travel, and some motion you should be able to determine it pretty
closely. Another way is to use two transmitters and fit a line
whose angle give vehicle orientation.

ARC International
4000 Chesswood Drive
Downsview, Ontario
Canada M3J 2B9
(416) 630-0200
					nivek


---

aka: Kevin Dowling			Carnegie Mellon University
tel: (412) 268-8830			The Robotics Institute
adr: nivek@rover.ri.cmu.edu		Pittsburgh, PA 15213


